Strands Lab AWS

CLI-based multi-agent assistant built with Strands on AWS Bedrock.

What It Does

recording_trimmed_optimized.mp4

• Automatically routes user requests across:
  • football_agent (22 football tools)
  • jobs_agent (resume + job-finder swarm)
  • chat_agent (general conversation)
• Streams output with thinking (gray) vs final answer (normal)
• Persists sessions per generated user ID
• Includes first-run .env bootstrap prompting

Requirements

• Python >=3.11
• uv
• AWS Bedrock access for the model in your configured region

Install

uv sync
uv run install-browsers  # Install Playwright Chromium for football agent scraping

Run

uv run cli.py

On first run, if .env is missing/incomplete, the app prompts for required keys and writes .env:

• AWS_REGION
• AWS_ACCESS_KEY_ID
• AWS_SECRET_ACCESS_KEY
• BEDROCK_MODEL_ID

CLI Commands

• /help show commands + internal agent overview
• /clear clear terminal
• /exit exit CLI
  also supports: /quit, /q, :q, :wq

Resume Loading Behavior

When resume analysis is requested, load_resume:

1. Prompts for pasted/dragged resume path
2. If sample exists, allows Enter to use sample
3. Loads .pdf or .txt

Default sample path lookup:

• /resumes/sample/hla.sample.pdf
• <repo>/resumes/sample/hla.sample.pdf

Project Layout

• cli.py - CLI entrypoint and prompt loop
• cli_config.py - CLI config/styles/command metadata/startup banner
• cli_stream_renderer.py - event stream rendering logic
• env_bootstrap.py - first-run env setup helper
• services/orchestrator.py - graph orchestration and routing
• services/fotmob.py - football tools
• services/job_swarm.py - job swarm and resume/job tooling